Thanks for the reply.
After considering the problem further, the problem I would be trying to solve with such a rule is disclosing human creativity.
Reviewing the linked quality guidelines, it appears they're worried about legal implications -- presumably around abuse of fair use, which would imply that an "ethically trained" model MIGHT squeak by moderation's concerns, but wouldn't do anything to assure users worried about shortcuts.
Viewing post in Generative AI Disclosure tagging
My guess it's mostly about the transitive propagation of the "has gen ai content" attribute.
If you make an asset with gen ai, you know what you did.
If you obtain a premade asset and include it in your game, you do not - unless it was properly advertised as such.
And since a lot of people are concerned about gen ai usage for a lot of reasons, they want clear labeling. And there are even some laws that try regulating that ai content and ai bots need to be labeled as such. What's happening in games is tame compared to all the crap that is happening with ai at other places.